48 <strong>Bt</strong> <strong>Brinjal</strong>: <strong>The</strong> <strong>GEAC</strong> <strong>environmental</strong> <strong>risk</strong> <strong>assessment</strong>control for brinjal. This is a complex question that will require some effort to gain clarity.Finding 23. <strong>The</strong> agronomic performance <strong>and</strong> efficacy experiments for hybrid <strong>Bt</strong> brinjal are designed for large-scalecommercial brinjal production systems, <strong>and</strong> do not reflect <strong>the</strong> production systems used by small-scale resource-poorfarmers. <strong>The</strong> data are probably appropriate for about 4% <strong>of</strong> brinjal production in India.<strong>The</strong> multi-site trials (MST, Dossier, volume 6) <strong>and</strong> <strong>the</strong> large-scale trials (LST, Supplemental Materials) are not designed toreflect <strong>the</strong> diversity <strong>of</strong> production systems in India. Specifically, <strong>the</strong>se experiments mimic <strong>the</strong> production systems <strong>of</strong> large-scalecommercial brinjal producers who already purchase hybrid seed <strong>and</strong> do not reflect <strong>the</strong> production systems used by small-scaleresource-poor farmers.A. No trials were conducted where <strong>the</strong> most brinjal is grown. West Bengal, Orissa, <strong>and</strong> Bihar produce 61% <strong>of</strong> <strong>the</strong> brinjalin India. <strong>Bt</strong> brinjal has been banned from <strong>the</strong>se states, but because agronomic performance varies with geographic locality,<strong>the</strong> results from <strong>the</strong> MST <strong>and</strong> LST can apply at most to 39% <strong>of</strong> <strong>the</strong> brinjal production in India.B. Comparisons are made among a <strong>Bt</strong> brinjal hybrid, a genetically similar non-<strong>Bt</strong> brinjal hybrid, a commercial hybrid withsimilar fruit morphology, <strong>and</strong> a local variety with similar fruit morphology (in year 2 <strong>of</strong> <strong>the</strong> MST). <strong>The</strong> local variety used in<strong>the</strong> MST experiments is not necessarily <strong>the</strong> most common or preferred local variety. Thus, <strong>the</strong> brinjal varieties are <strong>of</strong>ten notsuitable for estimating pr<strong>of</strong>itability <strong>of</strong> small-scale resource poor farmers, who grow local varieties.C. Agricultural inputs in <strong>the</strong> MST <strong>and</strong> LST are those recommended for large-scale commercial production. Small-scaleresource-poor farmers typically use saved seed, manures <strong>and</strong> composts for fertilisers <strong>and</strong> locally produced insecticides <strong>and</strong>h<strong>and</strong>-rouging for pest control. Thus, <strong>the</strong> production system used are appropriate for large-scale commercial farmers, but notfor small-scale resource-poor farmers. On average, about 10% <strong>of</strong> <strong>the</strong> brinjal producers outside <strong>of</strong> West Bengal, Orissa <strong>and</strong>Bihar are large-scale commercial producers (APEDA 2009). So <strong>the</strong> MST <strong>and</strong> LST data are appropriate only for about 4% <strong>of</strong><strong>the</strong> brinjal producers in India; <strong>the</strong>se are producers that are outside <strong>the</strong>se states <strong>and</strong> produce on large-scale (100 x 0.39 x 0.10).D. <strong>The</strong> LST field trials were probably conducted on soils representative <strong>of</strong> intensive, large-scale, commercial production.Availability <strong>of</strong> potash, a critical macro nutrient for brinjal depends on many factors. Under intensive, large-scale, commercialbrinjal production, soil organic matter content (SOM) can drop to 0.3-0.4%. When SOM is this low, brinjal yield potentialdepends primarily on <strong>the</strong> levels <strong>of</strong> external input. SOM for many small-scale resource-poor farmers is typically muchhigher than this, <strong>and</strong> under <strong>the</strong>se conditions, brinjal yield is a complicated function <strong>of</strong> soil quality <strong>and</strong> external inputs.Consequently, <strong>the</strong> agronomic yield data probably do not represent production systems <strong>of</strong> small-scale resource-poor farmers.Although soil pH <strong>and</strong> SOM content are not reported for <strong>the</strong> LST field trials, <strong>the</strong> trials were conducted on experimental farmsthat were likely to reflect st<strong>and</strong>ard, commercial production systemswith low SOM.Figure 3. Measure <strong>of</strong> yield gap between yields observedE. <strong>Brinjal</strong> yield loss from BFSB is different for large-scale in experimental trials <strong>and</strong> average yields on farm.commercial producers <strong>and</strong> small-scale resource-poor farmers (Table1 in Context <strong>and</strong> Need section). <strong>The</strong> MST <strong>and</strong> LST assess yieldloss <strong>the</strong> way large-scale commercial producers would assess it.Small-scale resource-poor farmers sell partially infested fruits on <strong>the</strong>market <strong>and</strong> consume <strong>the</strong> unsold remainder. Less damaged fruitscan be marketed at a good price because BFSB damage symptomsindicate to some consumers that reduced amounts <strong>of</strong> pesticideshave been used.F. A critical economic value <strong>of</strong> small-scale resource-poorfarmers is not considered. For <strong>the</strong>se farmers, brinjal shelf life iscritical for determining <strong>the</strong> economic value <strong>of</strong> <strong>the</strong> brinjal crop.Under small-scale production practices, brinjal typically has amuch longer shelf life than brinjal produced under large-scaleproduction practices. <strong>The</strong> longer shelf life means brinjal retains its
Socioeconomic Analysis 49economic value for a longer period <strong>of</strong> time, <strong>and</strong>small-scale resource-poor farmers may distribute<strong>the</strong>ir supply over time to obtain higher <strong>and</strong>/or steadier income. In this way, brinjal isessential for <strong>the</strong> economic security <strong>of</strong> small-scaleresource-poor farmers.Large-scale commercial growers compriseabout 10% <strong>of</strong> all brinjal producers outside <strong>of</strong> <strong>the</strong>main production areas in West Bengal, Orissa,<strong>and</strong> Bihar. <strong>The</strong>refore, <strong>the</strong> pr<strong>of</strong>itability analysispresented by EC-II is appropriate for only about4% <strong>of</strong> all <strong>the</strong> brinjal producers in India (0.1 x(1 – 0.61)).Despite all <strong>of</strong> <strong>the</strong> limitations, a narrowutilitarian may still suggest that <strong>the</strong> MST <strong>and</strong>LST data can be used to estimate pr<strong>of</strong>itability<strong>of</strong> <strong>Bt</strong> brinjal for farmers in India. This is awidely-shared perspective <strong>and</strong> is not as simpleto refute as <strong>the</strong> previous argument might makeevident. For example, limitation A is alreadytaken into account in <strong>the</strong> calculation in <strong>the</strong>previous paragraph, <strong>and</strong> Table 1 (Context <strong>and</strong>Need section) indicates that limitation E mightbe addressed by discounting <strong>the</strong> MST <strong>and</strong> LSTresults by 2/3rds. In a similar way, it couldbe argued that limitations B, C, <strong>and</strong> D can beaddressed by discounting <strong>the</strong> MST <strong>and</strong> LSTby some additional unknown X% to take intoaccount those data limitations. For lack <strong>of</strong> abetter value, a narrow utilitarian might perhapssuggest that 50% would suffice. <strong>The</strong>refore, <strong>the</strong>pr<strong>of</strong>itability for small-scale resource-poor farmersmight be 1/6 th that estimated by EC-II. However,this calculation does not take into accountlimitation F or <strong>the</strong> issue <strong>of</strong> yield gaps, which iscovered next.Finding 24. Yield gaps are prevalentbetween experimentally estimated yield<strong>and</strong> average farmer yield. <strong>The</strong> yield benefit<strong>of</strong> hybrid <strong>Bt</strong> brinjal estimated from <strong>the</strong>controlled MST <strong>and</strong> LST experiments shouldbe multiplied by 0.54 to estimate <strong>the</strong> yieldbenefit for <strong>the</strong> average large-scale commercialfarmer. This also reduces <strong>the</strong> estimatedbenefit to small-scale resource-poor framers.Box 12<strong>Brinjal</strong> Yield Gap: Overestimation <strong>of</strong> Yield Benefit <strong>of</strong> <strong>Bt</strong> <strong>Brinjal</strong>Because <strong>of</strong> <strong>the</strong> existence <strong>of</strong> yield gaps between yields measured underexperimental conditions at agricultural colleges <strong>and</strong> experiment stationsversus <strong>the</strong> yields obtained on farmer fields (Fig. 3), <strong>the</strong> yield benefit <strong>of</strong> <strong>Bt</strong>brinjal has been overestimated. This can be demonstrated as follows.Suppose Y <strong>Bt</strong>is experimental <strong>Bt</strong> brinjal yield <strong>and</strong> Y n<strong>Bt</strong>is <strong>the</strong> experimentalnon-<strong>Bt</strong> brinjal yield. <strong>The</strong> yield benefit <strong>of</strong> <strong>Bt</strong> brinjal is calculated in <strong>the</strong> MST<strong>and</strong> LST as Y <strong>Bt</strong>– Y n<strong>Bt</strong>. Because yields determined from experimentsoverestimate average farmer yields by <strong>the</strong> proportional yield gap, PYG E, bothY <strong>Bt</strong><strong>and</strong> Y n<strong>Bt</strong>should be multiplied by PYG Eto estimate <strong>the</strong> average farmeryields for <strong>the</strong> same experimental treatments. Thus, PYGE x Y <strong>Bt</strong><strong>and</strong> PYG ExY n<strong>Bt</strong>estimate average farmer yields for <strong>the</strong> two experimental treatments.<strong>The</strong> yield benefit <strong>of</strong> <strong>Bt</strong> brinjal for <strong>the</strong> average farmer is consequently PYG EY <strong>Bt</strong>– PYG EY n<strong>Bt</strong>, which can be rewritten as PYG E(Y <strong>Bt</strong>– Y n<strong>Bt</strong>). Thus, <strong>the</strong> yieldbenefit for <strong>Bt</strong> brinjal was overestimated in <strong>the</strong> MST <strong>and</strong> LST by <strong>the</strong>proportion given by PYG E.<strong>The</strong> proportional yield gap for rice <strong>and</strong> maize can be estimatedfrom data published in Lobell et al. (2009). <strong>The</strong> majority <strong>of</strong> data on rice isfrom India. <strong>The</strong> data on maize is mostly from <strong>the</strong> US, Latin America <strong>and</strong>Africa. <strong>The</strong> distribution <strong>of</strong> estimated proportional yield gaps is shown in <strong>the</strong>figure below.<strong>The</strong> average proportional yield gap for <strong>the</strong>se data is 54%.Considerable research efforts are expended to reduce <strong>the</strong> yieldgap (= bring <strong>the</strong> PYG up to 1.0) <strong>and</strong> bring average farmer yields up to <strong>the</strong>levels obtainable on experiment stations. To do this, it is essential tounderst<strong>and</strong> <strong>the</strong> underlying causes <strong>of</strong> <strong>the</strong> yield gaps, which are many <strong>and</strong>varied. <strong>The</strong> proportional yield gaps for US maize in Nebraska, <strong>the</strong> heart <strong>of</strong><strong>the</strong> Corn Belt were 40% <strong>and</strong> 56%. <strong>The</strong>se values are very close to <strong>the</strong> meanfor all countries, including many developing countries, so <strong>the</strong> high technologyused in <strong>the</strong> US does not eliminate yield gaps; likewise, <strong>Bt</strong> brinjal isunlikely to substantially reduce <strong>the</strong> brinjal yield gap. Probably <strong>the</strong> mostsignificant factors contributing to yield gaps are a class <strong>of</strong> factors related tolocal variation or field-specific factors (Lobell et al. 2009). <strong>The</strong>se includefarmer variation in management <strong>and</strong> wea<strong>the</strong>r <strong>and</strong> soil conditions. Considerableeffort must be expended to reduce yield gaps (= raise proportionalyield gaps to 1.0). This implies for <strong>Bt</strong> brinjal that simply making <strong>the</strong>technology available to farmers will not result in <strong>the</strong> dramatic yieldincreases promised in EC-II. Instead, Mahyco <strong>and</strong> <strong>the</strong> Indian governmentmust invest considerable amounts in outreach efforts so any yield benefitscan be realized. Such an investment, <strong>of</strong> course, reduces <strong>the</strong> socialeconomic surplus that <strong>Bt</strong> brinjal might generate <strong>and</strong> <strong>the</strong>reby reducing itssocial value.